Matthew Winn

I raise the ai in bicep since the s is tacked on to the next syllable, leaving "bi" as open... perhaps your ai in bicycle is raised because you sorta want to say "bike"?

Check your email :)

The Eras of Taylor Swift's Changing Dialect - Across Acoustics Longitudinal studies of how an individual's accent changes over the course of their life are hard to come by. Fortunately, Taylor Swift's decade-plus career-- and the numerous interviews she's given o...

Podcast episode about the acoustic analysis of Taylor Swift's changing vowels! featuring Miski Mohamed and myself

More info on the study available here!
sites.google.com/umn.edu/list...

including: summary of the results, why we chose to analyze Taylor Swift, the scientific and social value, how you can make your own measurements, and the greater context of what our lab does

cartoon showing a speech sound being used for acoustic analysis to show changes in vowel articulation in different cities

Taylor Swift changed her accent when living in different cities. Miski Mohamed and I analyzed the acoustics of her vowel articulation, showing key features of Southern accent during Taylor’s time in Nashville, and lowered voice pitch during her time in New York City
pubs.aip.org/asa/jasa/art...

Using Ant-Man’s voice as a lesson on voice acoustics and vocal tract length
www.youtube.com/watch?v=m0qx...

an illustration of speech displayed as a waveform (top) and a spectrogram (bottom). The original speech is on the left, and it is transformed into three different kinds of vocoded speech to the right (sinewave carriers, noise-band carriers, and harmonic-complex carriers)

Want to vocode speech? You can use our paper and code! This is a technique used to control spectral resolution and other factors that are important for understanding cochlear implants
github.com/ListenLab/Vo...
pubs.aip.org/asa/jasa/art...

Intro to Speech Acoustics - YouTube A series of videos on speech acoustics, for students in speech & hearing sciences, linguistics, psychology, etc.

Preparing to teach phonetics / acoustics in the fall? Here are some class videos that might be helpful for you!
www.youtube.com/playlist?lis...

Few people know that one of the many times I asked you for coding help, you pointed your finger to the sky like a superhero about to take flight, and exclaimed "to the source code!"

citation: Greenwood, D. (1990). "A cochlear frequency-position function for several species – 29 years later". Journal of the Acoustical Society of America. 87 (6): 2592–2605.

substitute continuous vs. log10 scaling for the axis in the code as you please!

ggplot(data.frame(pos = seq(0,32,1)))+
aes(x = pos, y = mm2freq(pos))+
geom_point()+
scale_x_continuous(breaks = seq(0,32,4),
name = "Basilar membrane position (mm from apex)")+
scale_y_log10(
name = "Frequency (Hz)",
breaks = round(500*2^(seq(-5,5,1))))

a graph showing basilar membrane position (in terms of millimeter distance from the apex) on the x axis, ranging from 0 to 32, and sound frequency log-scaled on the y axis ranging from 31 to 16000

a graph showing basilar membrane position (in terms of millimeter distance from the apex) on the x axis, ranging from 0 to 32, and sound frequency linearly-scaled on the y axis ranging from 31 to 16000

or, in reverse, find the frequency that lives on a specific spot on the basilar membrane

mm2freq <- function(position=NULL, A = 165.4, a = 2.1, length = 35, k = 0.88){
return(A*((10^((a*position)/length))-k))}

ggplot(data.frame(freq=10^(seq(1.4,4.3,0.1))))+
aes(x = freq, y = freq2mm(freq))+
geom_point()+
scale_y_continuous(breaks=seq(0,32,4),
name = "Basilar membrane position (mm from apex)")+
scale_x_log10(
name = "Frequency (Hz)",
breaks = round(500*2^(seq(-5,5,1))))

a graph showing sound frequency log-scaled on the x axis ranging from 31 to 16000, and basilar membrane position (in terms of millimeter distance from the apex) on the y axis, ranging from 0 to 32

a graph showing sound frequency linearly-scaled on the x axis ranging from 31 to 16000, and basilar membrane position (in terms of millimeter distance from the apex) on the y axis, ranging from 0 to 32

use R to pinpoint where a sound frequency falls on the basilar membrane in the cochlea

freq2mm <- function(freq=NULL, A = 165.4, a = 2.1, length = 35, k = 0.88){
return(log10((freq/A)+k)*length/a)}

need more real-life examples like this to demonstrate audio concepts

with geom_box(), every plot is a box plot

Interesting (although not too surprising) that data viz comprehesion uses two-stage process: 1st fast initial pass based on prior knowledge and assumptions about how graphs should behave, 2nd detailed orientated pass only if suggestion something is misleading.

doi.org/10.1007/s106...
#AcademicSky

Look, I love these sites & I get how politics works & why they're doing it, but every single one makes exactly the same error universities have been making for years:

Research DOES NOT JUST MEAN STEM.

Look around. If you think we also don't need social science & humanities research, look harder.

1. Today the NIH director issued a new directive slashing overhead rates to 15%.

I want to provide some context on what that means and why it matters.

grants.nih.gov/grants/guide...

then we climb up the vowel space and land on ʊ where it belongs!

Wooder is definitely caught when I'm outside the Philly region
Must... retain ... rounding!

(2/2)... people with cochlear implants show more carryover effort from one utterance to the next, which would impair flowing conversation. Importantly, the amount of carryover effort was not predictable just based on the listener’s percent-correct score!

Mentally fixing a word that you missed can lead to missing the next thing the talker says. This is a common problem for people who have hearing loss, but is ignored by audiological testing. Check out our new open-access paper on this topic here!
journals.lww.com/ear-hearing/...

chart showing measurements of fundamental frequency (pitch) and formant values for 34 iterations of the word "guilty" spoken by Ari Melber when reading the verdict

When a talker repeats the same vowel sequence 34 times in quick succession, a good phonetician simply can’t let that opportunity pass.
I present to you: the definitive acoustic analysis of Ari Melber's pronunciation of "guilty"

Academic mentorship is a skill that has to be practiced intentionally - it doesn't just happen automatically. Auditory friends, come join us in this chat on how to be a good mentor

My preferred guiding principle is to make it clear what your scientific identity is. Very often a person thinks their job is to be good at lots of things, but they are more rewarded for clearly being THE go-to person for a specialty.

Need plot
Need bike plot
Need bike plots
Need to build bike plots line by line

you found that two groups of children have different means on X, not that X can distinguish between two groups of children. keep it clear and boring.

Need more plots, all the time

this one is super well done, by @willngiam.bsky.social :
williamngiam.github.io/blog/my_peer...